comment
This commit is contained in:
parent
d9d87fdbc9
commit
f6f2ce371c
1 changed files with 48 additions and 0 deletions
|
@ -0,0 +1,48 @@
|
||||||
|
[[!comment format=mdwn
|
||||||
|
username="joey"
|
||||||
|
subject="""comment 1"""
|
||||||
|
date="2015-11-10T15:36:15Z"
|
||||||
|
content="""
|
||||||
|
Well, it's definitely an oversight that it still fails like that when
|
||||||
|
~/.ssh can't be created. I fixed that earlier for the ~/.ssh/git-annex-shell
|
||||||
|
creation but missed doing the same for the ~/.ssh/git-annex-wrapper
|
||||||
|
creation. Now fixed.
|
||||||
|
|
||||||
|
----
|
||||||
|
|
||||||
|
It might have been better to put these files someplace else like
|
||||||
|
~/.config/git-annex (although that's only a standard location on Linux and
|
||||||
|
this is also done on OSX) or ~/.ssh/git-annex/ (which the assistant puts
|
||||||
|
keys in so I think would be a better choice). Transitioning to a new
|
||||||
|
location now would require writing them to both locations long enough for
|
||||||
|
all current git-annex assistant users to upgrade. I don't know how long
|
||||||
|
that would be. The utility in moving the files down 1 directory level does
|
||||||
|
not seem worth the disruption.
|
||||||
|
|
||||||
|
I also sympathize with not wanting those when not using the assistant.
|
||||||
|
But, this is intended for users who are using the standalone builds to get
|
||||||
|
git-annex installed on a server to use with the assistant. People have no
|
||||||
|
end of difficulty getting git-annex-shell into PATH (bash actively makes
|
||||||
|
this exceedingly difficult by refusing to read any config files for
|
||||||
|
noninteractive shells), and so these files avoid a particularly newbie
|
||||||
|
class of users struggling with that problem.
|
||||||
|
|
||||||
|
So, the idea is they unpack the tarball somewhere, and run runshell, and
|
||||||
|
can see git-annex working on the server, and then the assistant magically
|
||||||
|
can talk to that git-annex server without them needing to worry about how.
|
||||||
|
This is why runshell is standing in for an installation process.
|
||||||
|
|
||||||
|
Adding another, more explicit installation step for this class of users
|
||||||
|
is just going to reduce the number who get git-annex installed on servers,
|
||||||
|
I fear.
|
||||||
|
|
||||||
|
---
|
||||||
|
|
||||||
|
But, /usr/lib/git-annex.linux/ .. this is the git-annex-standalone.deb
|
||||||
|
isn't it? Since debs are a much better way to get things into PATH
|
||||||
|
and get dependencies installed etc, there's certianly no reason for the
|
||||||
|
runshell in the standalone deb to do this installation.
|
||||||
|
So, I've fixed that..
|
||||||
|
|
||||||
|
Enough to close bug?
|
||||||
|
"""]]
|
Loading…
Add table
Add a link
Reference in a new issue